skills/youtube-script/references/commentary-patterns.md

# Commentary / Essay — Scriptwriting Craft Patterns

**Corpus:** Derived from ~80 top-performing scripts across 4 leading channels in the genre (sources anonymized for publication). Deep reads: top 4 per channel (manual/whisper captions preferred); hook reads (first ~40 lines) on all remaining. Where sources are auto-captions (unpunctuated), sentence-level rhythm claims are inferred from spoken cadence, not punctuation.

**Archetype key (used throughout):**
- **The affable critic** — comedic commentary; found-footage reaction with heavy sketch elements and a self-deprecating persona.
- **The absurdist investigator** — comedic commentary; discovery/experiment formats, character-voice hypotheticals, elaborate audience lore.
- **The deadpan satirist** — short deadpan analytical comedy over simple drawings; parody-explainer format.
- **The close-reading essayist** — earnest video essays on film, art, and rhetoric; no jokes required.

All quoted lines below are invented illustrations of each formula, not lines from any real script.

---

## 1. Voice & register

**Person and tense.** All four archetypes are first-person, present tense. Clips are narrated in the *narrative present* ("So then the guy walks straight into the manager's office…"), which keeps found footage feeling live rather than recapped. The close-reading essayist analyzes in the present too ("The speech leans on tiny sentences"): art is treated as an ongoing event you're both looking at right now.

**Contractions and diction.** Contractions everywhere; profanity is casual seasoning in the two comedic commentators, absent in the essayist, occasional and deadpan in the satirist ("Unsalted crackers. Absolutely shameless."). Vocabulary is spoken-register even when the idea is sophisticated — the essayist will say "honestly, it's kind of gorgeous" rather than reach for academic phrasing, then immediately back it with hard data.

**Sentence rhythm.**
- **The comedic commentators:** short spoken bursts, frequent mid-sentence pivots and self-interruptions ("I was fully ready to wrap up right here, but then…"). Questions are constant — the host asks the clip questions the audience is already thinking ("Whose mannequin is this?").
- **The deadpan satirist:** long, grammatically calm compound sentences delivered flat, engineered so the joke is the final clause. The sentence *sounds* like advice until the last five words detonate it.
- **The close-reading essayist:** medium-length periodic sentences that resolve onto thesis words; he even analyzes this pattern in others ("landing every sentence on its heaviest word") and practices it himself.

**Persona-to-audience relation: peer, never authority.** This is the category's defining register. The affable critic and the absurdist investigator watch the material *with* you and get confused *with* you ("I genuinely cannot follow what's happening here"). The deadpan satirist plays an authority — but a visibly unqualified one whose advice always collapses back onto himself; the deadpan expert is the joke. The close-reading essayist is the most authoritative, but he hedges nearly every claim through personal experience ("I think," "for me," "I honestly didn't expect this"), positioning himself as an enthusiast sharing a discovery, not a professor grading you. Both comedic archetypes maintain audience-as-in-group rituals: the absurdist investigator opens by greeting the audience by an invented collective name (imagine a channel that opens every video with "Greetings, Cadets"), and the affable critic addresses the entire audience as one single person (imagine "Hello, lone viewer") — the address itself is a running bit that makes the viewer a character.

---

## 2. Hook taxonomy

Every distinct opening move found in the first 30–60 seconds across the corpus:

**H1. Cold-open bit / in-media-res sketch.** Start inside a joke with zero context; the topic emerges from the bit. Formula: *[Act out a character/scene that obliquely sets up today's subject] → snap to normal register.* The affable critic opens one video in full costume as a medieval town crier announcing the subject; another opens with a fake mournful monologue about his own cancelled plans that pivots on "Oh good — the scandal-prone influencer is going on an arena tour!"; the absurdist investigator opens one video with a wordless prop gag before anyone speaks. Used by: the affable critic (heavily), the absurdist investigator.

**H2. Rotating fake show title.** The greeting ritual plus a one-off show name that pre-frames the topic. Formula: *"[Greeting], welcome back to another episode of [absurd show name that describes this video]."* Invented examples: "another episode of My Beautiful Soup Kingdom"; "welcome back to the focus group in my garage." Used by: the absurdist investigator (near-every video), the affable critic.

**H3. Artifact-first cold open.** Show the weird thing itself before any framing. Formula: *[Play/read the artifact raw] → host reacts → "has anyone else noticed…"* The affable critic opens one video by reading machine-generated ad copy verbatim; another dramatizes the ad interrupting his day; the close-reading essayist runs a viral political speech's full answer cold before a word of analysis. Used by: the affable critic, the absurdist investigator, the close-reading essayist (his default).

**H4. Discovery/origin-story hook.** How the material found the host — establishes communal stakes. Formula: *"There's been this [thing] going viral / people keep sending me [X], so I had to look into it."* Invented examples: "About four hundred of you tagged me in this"; "clips from this movie keep colonizing my feed." Used by: the absurdist investigator (signature), the affable critic.

**H5. Experiment-stakes hook.** Declare a costly, irreversible commitment up front. Formula: *"I'm going to [do expensive/painful thing] to answer [question]."* Invented examples: "I am going to spend TWENTY-FIVE THOUSAND DOLLARS on this"; "is every storage-auction show completely staged?"; "and yes — I'm doing it again." Used by: the absurdist investigator, the affable critic.

**H6. Verdict-first thesis hook.** Open with the conclusion, then earn it. Formula: *"[X] is bad. The jokes aren't funny… let me explain."* The affable critic opens a takedown of a daytime talk show reboot flatly: "this show is bad"; the absurdist investigator: "it's simply not good." Used by: the affable critic, the absurdist investigator, the close-reading essayist (positive polarity: "there's maybe no single frame more worth your attention than this one").

**H7. Deadpan definition hook.** Wikipedia-voice definition of an everyday phenomenon, played straight for exactly one sentence. Formula: *"[Topic] is essentially [dry textbook definition]…" → first undercut within ~20 seconds.* Invented examples: "a warning sign is essentially a small detail you notice early and ignore professionally…"; "small talk is essentially a verbal handshake nobody requested…" Used by: the deadpan satirist (signature).

**H8. Fake-service hook.** Promise genuinely useful help, then deliver anti-help. Formula: *"I know a lot of people struggle with [X], so I made this guide/quiz."* Invented example: "so I've put together a short assessment" — the identical shape recurs across his catalog. Used by: the deadpan satirist.

**H9. Question hook.** A single concrete question the whole video answers. Formula: *"Here's a question for [audience]: is it possible to [X]?"* Invented examples: "Is it possible to fix this divisive space romance with a single re-edit?"; "why does one editor control how I see everything?" Used by: the close-reading essayist.

**H10. Singular-object superlative hook.** One artifact, one extreme claim. Formula: *"Of all [category], the most [adjective] is this one."* Invented examples: "of every prop in this franchise, the one that unsettles me most is the puppet"; "there's one shot in this animated blockbuster that completely rewired how I watch." Used by: the close-reading essayist, the deadpan satirist (inverted, mock-superlatives).

**H11. Personal-confession hook.** Vulnerability as entry ticket. Formula: *"[Embarrassing autobiographical fact] which is why I can explain [topic]."* Invented examples: "I moved abroad knowing exactly zero words of the language"; the affable critic opens one video on losing money at a county-fair raffle: "genuinely one of the best days of my life." Used by: the deadpan satirist, the affable critic.

**H12. Pre-announcement open loop.** Housekeeping weaponized for retention. Formula: *"Stick around to the end for [announcement/reveal]."* The affable critic opens one video with exactly this before the sketch. Used by: the affable critic, the absurdist investigator (tour/announcement versions).

Note the common denominator: **within 30 seconds you always know the artifact and the host's stance toward it.** No channel opens with generalized throat-clearing.

---

## 3. Structural beat map

### Comedic commentary long-form (the affable critic, the absurdist investigator; 15–30 min)

| % of runtime | Beat |
|---|---|
| 0–3% | Cold-open bit / greeting ritual / fake show title (H1/H2) |
| 3–10% | Origin story + credentials ("that video did numbers I still don't understand") + promise of today's premise; optional announcement loop |
| ~10% | Hard turn: "Anyway, gang, let's open the box!" — an explicit spoken chapter break |
| 10–80% | **Clip-react loop** (see below), segments ordered by ascending absurdity |
| 40–60% | Mid-video escalation or thesis-deepening pivot: the affable critic stops a talk-show critique to essay on creator ego ("it's because famous internet people assume every skill transfers"); the absurdist investigator's ad campaign gets frozen mid-experiment — real events become act breaks |
| 80–90% | Verdict + sincere zoom-out: the one earnest paragraph (the affable critic on an influencer collective's effect on young fans; "What kind of world does this craft-hack channel actually want us living in?") |
| 90–100% | Subscribe lore, next-video teaser joke, recurring outro jingle |

**The clip-react loop** (the atomic unit, ~30–90s per cycle): play 5–20s of subject material → literal restatement of what happened ("Okay, several things are happening at once here. First…") → logic interrogation (why would X?) → escalate into an act-out or hypothetical sketch in character voices → tag → next clip. Reaction time roughly equals or exceeds clip time; the host never lets footage run past ~20 seconds uninterrupted.

### Deadpan analytical comedy (the deadpan satirist; 2–8 min)

Definition (0–10%) → announced taxonomy: tiers, stages, levels, steps ("entry-level excuses… mid-tier… elite tier"; a mock hierarchy diagram) → each item = setup-twist unit (~15–25s) → **final reversal that turns the lens on the speaker/viewer**: "and the elite-tier warning sign: you're the one everyone at the party is avoiding." The list scaffold *is* the act structure; the escalating tiers are the escalation logic; the self-implicating ending is the third act.

### Earnest video essay (the close-reading essayist; 6–9 min)

Cold artifact (0–10%) → thesis stated in one "I think" sentence by ~15% ("I thought it would be worth sitting with one answer from this speech") → context/background (15–30%) → evidence beats of increasing granularity (30–75%): quantification → technique naming → replay/close-read → external corroboration → **widen to the biggest human meaning** in the last 20% ("What's actually at stake here is your attention.") → direct-address closer ("do yourself a favor and read the transcript") → chatty personal outro.

**Interleaving rule across the category:** subject material and reaction alternate in short strata; analysis always *retroactively recasts* the material just shown (the essayist's ending-recontextualizes-montage move; the affable critic replaying a clip after decoding it). The audience should never watch anything twice for the same reason.

---

## 4. Comedy mechanics

**Setup-punchline spacing.** The deadpan satirist runs the tightest interval in the corpus: a reversal roughly every 15–25 seconds, with the twist positioned as the *final clause* of a calm sentence — call it **last-word detonation**: "Maybe the interview went great and they simply lost my number for six years." The neutral setup is never signposted as a joke; delivery stays flat through the punchline. The comedic commentators space bigger laughs 30–60s apart but fill the gaps with micro-jokes (act-outs, one-word tags).

**Misdirection lists.** Three legitimate items, then one absurd item in the same grammatical costume: when a subject's own list does this it gets flagged, but the satirist's own lists do it constantly ("You can meet people at work, through friends, at the gym… and that's genuinely the whole list."). Formula: *N sincere + 1 absurd, no tonal change.*

**Fake precision.** Deadpan numbers sell irony: "a 4 to 5.5 out of 10 is the statistically safest zone"; "the legendary form corridor." The absurdist investigator inverts it with real numbers played for absurdity ("$1,111.11 per day, per video").

**Callback architecture.** Three tiers:
1. *In-video plants:* the weirdest clip line is planted early, re-dropped unannounced 2–3 times, then labeled at the climax — the affable critic literally announces "Chandelier salad callback! Remember that from earlier?" and keeps hard-cutting to the same baffling reply. Labeling the callback is itself the joke.
2. *Series-level runners:* the fanbase-name lore, the singular-viewer greeting, end-screen jingles, "Thank you [name] for enabling notifications."
3. *Cross-channel runners:* the two comedic commentators maintain a long-running mock feud over which of them invented the format — callbacks as community currency.

**Absurdist escalation (the absurdist investigator's signature).** A three-stage engine: (1) take the clip's premise *literally* and find the contradiction ("The 'invisible' phone case is fully visible in every shot!"); (2) inflate it into a hypothetical scene performed in character voices (the security guard interrogating a sandwich for 45 minutes); (3) push to world-scale consequence ("and ten years from now every office building is just people pretending their phones don't exist…"). Literal → scene → dystopia.

**Self-deprecation placement (the affable critic's signature).** Deployed at three fixed points: the hook (bad-haircut jokes, "the only commentary channel filmed inside a closet"), at any moment of judgment against someone else (buying tickets to the influencer's arena tour is framed as his own idiocy: "some traitorous part of my brain… already bought two floor seats"), and inside the CTA. Function: it licenses the mockery — the critic pre-mocks himself, so the criticism reads as peer teasing, not superiority. He also fences ethical bounds explicitly before punching ("to be clear, this isn't some harmless eccentric we're piling on" — before documenting a subject's genuinely harmful behavior).

**Understatement & anti-correction.** The mock-generous misreading: "I'm sure that's a typo and they meant to charge three dollars." And the reverse — conceding the opponent's point sarcastically: "No wonder the network gave him a second season!"

**Act-outs as subtext-to-text.** Both comedic commentators voice what on-screen people are *really* saying ("Look, none of this matters, just point the camera at ME" — dubbing over a staged phone call). This is the single most-used joke shape in the two channels: restate the scene's hidden logic in the character's own mouth.

---

## 5. Analytical essay mechanics (the close-reading essayist)

**Thesis placement.** Cold-open artifact first, thesis second — always within the first 60–90 seconds and always framed as personal fascination, not decree: "What fascinates me about this speech is how hard it works to say almost nothing." The thesis is a *lens*, not a verdict; the verdict arrives at the end.

**Evidence sequencing (a consistent ladder):**
1. **Quantify** — count something concrete before interpreting anything: "This 340-word, ninety-second answer"; "48 shots in just under four minutes"; "that pause lasts a full five seconds." The counting earns trust for the interpretation.
2. **Name the technique** — attach a term of art (the Kuleshov effect, a split diopter, a readability formula — any general film or rhetoric term) with a fast inline definition, so the viewer leaves with a tool.
3. **Corroborate externally** — a study, an editor's process, a director's own words (a newspaper's syntax experiment; the editor's own account of stitching two takes into one).
4. **Interpret upward** — only now the meaning claim, still hedged with "I think."

**The "watch this moment again" move.** Play the moment cold → analyze → replay with the new frame so the audience *feels* the upgrade ("run it back frame by frame and something strange appears"). The essay's climax is often a recontextualization: an acclaimed foreign thriller's ending recasts its earlier montage entirely; a time-bending sci-fi drama's opening scene means the opposite once the chronology is understood. Structure lesson: sequence the essay so the *audience's own earlier perception* becomes evidence.

**Theme braiding.** Every observation must echo at least one other: "nothing introduced is allowed to go unused" — he states his own architecture out loud. Even his counterfactual method (re-editing a divisive space romance to prove a script point) is evidence-by-demonstration rather than assertion.

**Endings.** Widen from craft to the human stakes and pivot to second person: "But the next time that speech starts to sound persuasive, do yourself a favor and read the transcript."

---

## 6. Retention machinery

**Explicit escalation labels ("and it gets worse" pivots).** The single most transferable retention device in the comedic half: segments are joined with comparative superlatives — "And somehow it gets worse!"; "But the strangest part is still coming…"; "there's another scene that manages to top the elevator one"; "we've seen a lot of bad scenes today — this one retires the trophy." Each pivot promises the ladder still goes up.

**Real-time stakes resets.** In experiment formats, live setbacks become act breaks: "Overnight, the platform froze my entire ad campaign."; the affable critic's phone dying halfway through the arena show. The video's production problems are folded into the plot.

**Open loops.** Planted at the hook (front-loaded announcement teasers), and mid-video (award-show framing with a held reveal: "The winner will be revealed… right after these messages.").

**Meta-marked chapter turns.** Transitions are jokes about transitions: "and that noise you just heard was a segue"; "let's call a timeout here." The seams are shown, which keeps the peer register while still functioning as chapters.

**Micro-retention inside the loop.** Every clip is introduced as a question the host then answers ("What is this? Who approved it? And how did it end up on my desk?"). Confusion is performed first, resolved second.

**The deadpan satirist's alternative:** no chapters, no loops — retention is pure joke density plus a *known scaffold* (tiers/stages ascending), so the viewer always knows a higher tier is coming. The final tier is reliably the best twist, which trains completion.

**The close-reading essayist's alternative:** a curiosity gap held on a tiny concrete detail ("watch the left hand — it's a tiny thing, but it's everything") plus the promise that a replay will make you see it. His countdown structures (shot-by-shot, beat-by-beat through one joke or montage) give the same "known scaffold" effect as the satirist's tiers.

---

## 7. Sub-styles — and what transfers

**Comedic commentary (the affable critic, the absurdist investigator).** Found footage + persona. The continuity product is the *relationship* (fanbase name, singular-viewer greeting, runners); the jokes are generated from the material's internal logic. Weakness risk: dependency on the artifact's quality — mitigated by escalation ordering and act-outs.

**Deadpan analytical comedy (the deadpan satirist).** No footage, pure script over simple drawings. The genre being parodied (explainer/guide) supplies the structure; the comedy is register-contrast: textbook voice, confessional content. The persona is the ostensible expert who keeps revealing he's the case study.

**Earnest video essay (the close-reading essayist).** No jokes required. Authority is built by counting, naming, and replaying — never by credentials. The persona is the delighted close-reader.

**What transfers across all three (and to any channel):**
1. Artifact + stance inside 30 seconds.
2. Peer register — hedged claims, self-implication, shared discovery.
3. Known scaffold announced early (tiers, experiment, shot-count) so viewers can track progress.
4. Escalation ordering with labeled pivots.
5. The replay/recontextualize move — make earlier material mean more the second time.
6. One sincere widening near the end; irony alone doesn't close a video in this corpus.

---

## 8. Ending & outro handling

Monetization patterns are intentionally omitted: these skills target sponsor-free, at-scale production. What carries over is the genre's ending craft:

- The content verdict plus the one sincere zoom-out (80–90%) is the real ending; everything after it is a short in-persona outro.
- The outro may contain at most one plain next-video or subscribe line — best delivered as lore in persona ("subscribing automatically enrolls you in my imaginary homeowners association") — or nothing at all (the deadpan satirist stops dead).
- CTAs never precede the verdict or interrupt the clip-react loop; register never shifts to corporate — the persona survives to the last second.
- The final beat is identity, not an ask: a next-video teaser gag with an absurd premise, a recurring outro jingle, or an affectionate in-joke aimed at the viewer.

---

## 9. Anti-patterns — what these scripts avoid

- **Announce-y framing.** Nobody says "in today's video we'll cover three things." Structure is enacted, not previewed; when a preview exists it's a joke (fake show titles).
- **Neutral summary.** No stretch of plot/clip description exists without stance or joke attached; recap is always fused with reaction.
- **Letting footage breathe too long.** Clips rarely run past ~20 seconds without interruption; the host's voice is the product.
- **Unlicensed punching.** Criticism gets an ethics check first: the affable critic verifies his critique echoes affected groups ("don't take it from me — here's what the people involved actually said") and establishes real harm before full mockery. Targets are institutions, output, and behavior — not appearance or fans (fans are teased only from inside the in-group).
- **Manufactured hype.** The energy register they mock in others (an influencer collective's content described as "essentially a sequence of airhorns") is precisely what they never do; excitement is dry, specific, or self-mocking.
- **Hiding the seams.** Mistakes stay in and become bits ("Do I even cut this out at this point? No. It stays."). Over-polish would break the peer register.
- **Unsupported superlatives (the close-reading essayist).** Every "best/perfect" claim is immediately paid for with a count, a named technique, or a replay. No vibes-only analysis.
- **Explaining the joke** — except as a deliberate meta-bit (labeling a callback is a licensed exception).
- **Cliffhanger abuse.** Retention comes from density and escalation, not withheld information; the one "after these messages" gag parodies the practice rather than using it.
- **Register drift in the outro.** No script switches into corporate voice; the persona survives to the final second.
- **Front-loaded CTA longer than ~15 seconds.** When it happens it's self-aware and instantly cut off ("Okay. Commercial over. Here's the video.").

---

## 10. Formula card (standalone recipe)

**HOOK (0:00–0:30) — pick one, know your lane:**
- *Cold-open sketch:* act the premise for 10–20s, then snap to normal register.
- *Fake show title:* "[Greeting ritual], welcome back to another episode of [absurd show name describing this exact video]."
- *Artifact-first:* play/read the weird thing raw for 10s, then react.
- *Discovery:* "People kept sending me [X], so I finally looked into it."
- *Experiment:* "I [spent/did costly thing] to find out [question]."
- *Verdict-first:* "[X] is [bad/great]. Let me explain."
- *Deadpan definition (satirist-style):* textbook definition, twist inside 20 seconds.
- *Question/superlative (essay-style):* "Of all [category], this one [extreme claim]" — then count something.
By 0:30 the audience must know **the artifact and your stance.**

**BEAT TEMPLATE (20-min commentary):**
- 0–3% cold-open bit → 3–10% origin story + promise → hard spoken turn ("anyway, let's get into it")
- 10–80% clip-react loop: ≤20s clip → restate → interrogate logic → act-out/hypothetical → tag. Order segments by ascending absurdity; join with "and it gets worse" labels.
- 40–60% one pivot: a real setback, a thesis deepening, or a format twist.
- 80–90% verdict + one sincere zoom-out paragraph.
- 90–100% subscribe-as-lore → next-video teaser gag → outro runner.

**BEAT TEMPLATE (5-min deadpan explainer):** definition → announced tier/stage scaffold → 12–18 setup-twist units at ~15–25s each → final tier reverses onto the speaker/viewer → stop dead (no outro).

**BEAT TEMPLATE (8-min essay):** cold artifact → "I think" thesis by 15% → count something → name the technique → corroborate → replay the moment with new eyes → widen to human stakes → second-person closer → warm outro.

**REGISTER RULES:**
- First person, present tense, contractions, spoken diction. Peer, never lecturer.
- Hedge big claims through self ("I think," "for me"); state small claims with fake precision.
- Self-deprecate at the hook, before every punch, and in the CTA.
- Perform confusion before resolving it; voice on-screen subtext as dialogue.
- Keep seams visible; keep persona through the very last second.

**JOKE-DENSITY GUIDANCE:**
- Deadpan explainer: reversal every 15–25s; twist = final clause; never telegraph.
- Commentary: big laugh every 30–60s, micro-jokes (tags, act-outs, drop-ins) filling gaps; plant the weirdest found line early, drop it back 2–3×, label the last callback.
- Essay: zero required jokes, but one delight beat per minute (a count, a replay, a "did you see that?").
- Lists: N sincere items + 1 absurd, same grammar, same tone.
- Escalate in threes: literal contradiction → hypothetical scene → world-scale consequence.

---

*All quoted lines are invented illustrations of the formulas described; no verbatim material from any source script appears in this document.*